Vijay Shekhar Sharma is running Paytm on a simple bet: let machines do more of the thinking, but keep people where they can build growth. As AI moves deeper into the company's operations, Paytm is automating decisions across merchant acquisition, customer servicing and collections. At the same time, it is spending more on sales and service employees who take the business deeper into India. The result is a management model where AI is not replacing people - it is changing what people are expected to do.
The clearest example comes from Paytm's merchant business. During the company's Q1 FY27 earnings call, Sharma said its small-merchant acquisition operation is increasingly governed by an internally built AI agent that helps determine what field sales executives should do. Instead of asking a salesperson to figure out which merchant to approach and when, the machine provides the intelligence while the person executes on the ground. Haitong Securities highlighted this model in its post-results report, noting that Paytm's "in-house AI system now directs field sales executives for small-merchant acquisition, reducing cost-to-acquire even as absolute sales/marketing spend keeps rising."
Building AI for India, in-house
Sharma's larger argument is that AI alone may not stay a durable edge. "Basically, AI is a distribution business," he said on the call, arguing that what will ultimately differentiate companies is how many customers they have, their quality and how effectively they can monetise them. "Customer, customer quality, monetization and monetization ability, these are the factors that multiply once you add the power of AI," he said. That explains why Paytm is using AI to optimise costs while continuing to invest aggressively in consumer acquisition, merchant expansion and financial services.
The numbers show this distinction. Paytm's revenue rose 28% year-on-year to ₹2,448 crore in the June quarter, while EBITDA increased 182% to ₹203 crore. Total indirect expenses grew just 6%. Yet sales and service employee costs increased 27%, as Paytm expanded its presence in tier-2 and tier-3 cities. In contrast, the cost of building the platform declined 3%, while software and cloud expenses fell 5%, even as the company continued investing in AI.
Behind those efficiencies is increasingly home-grown technology. Sharma said Paytm took "a 200 billion parameters model, optimized it to a 4 billion parameters model made for Indian languages," and deployed it on its own infrastructure for functions including collection, retention and customer revisit. "So we remove the cost of the call center. We remove the cost which otherwise would have been bought from outside," he said, before adding: "Now, this is magic." Paytm now plans to offer some of these internally developed AI capabilities to outside businesses as a new revenue stream.
Every rupee, employee and machine must drive growth
The efficiency has not gone unnoticed by analysts. Haitong Securities said management "reiterated higher confidence in reaching the 15%-20% EBITDA margin band in the medium term," adding that this could now be achieved sooner than earlier thought, supported by revenue growing faster than indirect costs and AI-led productivity.
For Sharma, the philosophy extends beyond technology. It is increasingly about deciding where every rupee, employee and machine earns its place inside the organisation. Near the end of the earnings call, he put that test simply: "Either you give me growth or bottom line, everything else is, thank you so much."
That line may ultimately explain Paytm's "machine-first" philosophy better than any AI jargon. Machines are being asked to make the organisation faster and more productive. People are being concentrated where Paytm believes they can build distribution, customers and revenue. And Sharma appears intent on directing both toward where they can build the business.
Why this matters for sales professionals
Paytm's model points to a concrete shift for sales roles: AI is taking over the targeting and timing decisions, while people handle the relationship work on the ground. Sales executives who learn to act on AI-driven direction - rather than relying on their own intuition about which accounts to chase - are the ones who will keep their jobs as these systems roll out. For sales professionals, the practical takeaway is to start building familiarity with how AI agents prioritise leads and route activity, because that capability is becoming a baseline expectation, not a differentiator. The AI Learning Path for Sales Representatives covers exactly this kind of workflow, and the broader AI for Sales resources offer a starting point for understanding how these tools change daily execution.
Your membership also unlocks:
